Mining raw GPS readings for deep profiling context
This study explores the utilization of raw GPS data combined with deep learning techniques for location context profiling in urban environments. Focusing solely on IMU data extracted from raw GPS readings, our research aims to assess the feasibility of this approach in capturing location contexts wi...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/175294 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | This study explores the utilization of raw GPS data combined with deep learning techniques for location context profiling in urban environments. Focusing solely on IMU data extracted from raw GPS readings, our research aims to assess the feasibility of this approach in capturing location contexts without explicitly considering spatial or temporal dynamics. By training Convolutional Neural Networks (CNNs) on IMU data, our study demonstrates promising results, achieving high-test accuracies across various urban settings. Despite the absence of explicit analysis of spatial and temporal dynamics, our findings highlight the potential of leveraging raw GPS data for accurate and context-aware localization. Future research avenues may address challenges such as domain shifts and generalization issues, while exploring additional data sources to further enhance localization systems' accuracy and reliability. This research underscores the significance of accurate localization in facilitating navigation, supporting location-based services, and advancing smart city initiatives in urban environments. |
|---|